Question

1. How is processor performance currently improved after we hit the processor power wall?

2. How is CPU benchmarked based on the execution time as suggested by SPEC?

3. How is CPU benchmarked based on the power consumption as suggested by SPEC?

Answer 1

1.Making computers faster suggests that relying on the central method unit (CPU) however ever before.

The Central method Unit (CPU)–the half that has made public the performance of your computer for many years–has hit a wall.

In fact, the next-generation of CPUs, also as Intel’s forthcoming Sandy Bridge processor, have to be compelled to have an effect on multiple walls–a memory bottleneck (the metric of the channel between the pc hardware and a computer’s memory); the instruction level correspondence (ILP) wall (the accessibility of enough distinct parallel directions for a multi-core chip) and additionally the facility wall .

Of the 3, the ability wall is presently arguably the method limit of the ability of the modern element. As CPUs became extra capable, their energy consumption and heat production has grownup apace. It’s a tangle so tenacious that chip manufacturers square measure forced to create “systems on a chip”–conurbations of smaller, specialised processors. These systems unit so sprawling and varied that they’ve caused long-time trade observers like Linley Gwennap of small chip report back to question whether or not or not the initial definition of a element even applies to today’s chips.

In emotional Sandy Bridge, Gwennap observes, Intel has little to tout in terms of improved element performance:

Sure, they found several places to nip and tuck, studying some p.c in performance here and there, however it's laborious to enhance a extremely out-of-order four-issue central processor that already has the world’s best branch prediction.

Instead, Intel is touting the chips’ new integrated graphics capabilities and improved video handling, each of that square measure accomplished with elements of the chip dedicated to those tasks–not the central processor instelf, which might be forced to handle them in package and within the method use a far larger proportion of the chip’s power and warmth budget.

And what of all-purpose computing tasks? Gennap explains that here, paradoxically, the key to subjugation the facility wall isn’t a lot of power–it’s less. Fewer watts per instruction suggests that a lot of directions per second in an exceedingly chip that's already running as hot because it probably can:

The changes Intel did create were a lot of typically concerning power than performance. the rationale is that Intel’s processors square measure against the facility wall. within the recent days, the goal was to squeeze a lot of Mc out of the pipeline.

Today’s CPUs have Mc to burn however square measure throttled by the number of warmth that the system will pull out. scale back the central processor power by 100% and you'll push the clock speed up to compensate, turning power into performance gains. Most central processor style groups square measure currently a lot of centered on the facility budget than on the temporal arrangement budget.

This means that, a minimum of with this generation of chips, Intel is innovating anyplace however within the central processor itself.

2. In computing, a benchmark is that the act of running a bug, a group of programs, or alternative operations, so as to assess the relative performance of associate object, unremarkably by running variety of ordinary tests and trials against it. The term 'benchmark' is additionally principally used for the needs of in an elaborate way designed benchmarking programs themselves.

Benchmarking is sometimes related to assessing performance characteristics of element, for instance, the floating purpose operation performance of a central processor, however there square measure circumstances once the technique is additionally applicable to package. Package benchmarks square measure, for instance, run against compilers or direction systems.

Benchmarks give a way of scrutiny the performance of assorted subsystems across totally different chip/system architectures.

Test suites square measure a sort of system supposed to assess the correctness of package.

As pc design advanced, it became harder to match the performance of assorted pc systems just by viewing their specifications. Therefore, tests were developed that allowed comparison of various architectures. for instance, Pentium four processors usually operate at a better clock frequency than Athlon XP processors, that doesn't essentially translate to a lot of procedure power. A slower processor, with respect to clock frequency, might perform also as a processor operative at a better frequency. See BogoMips and also the Mc story.

Benchmarks square measure designed to mimic a selected form of employment on a part or system. Artificial benchmarks do that by specially created programs that impose the employment on the part. Application benchmarks run real-world programs on the system. whereas application benchmarks typically provides a far better live of real-world performance on a given system, artificial benchmarks square measure helpful for testing individual elements, sort of a magnetic disc or networking device.

Benchmarks square measure notably vital in central processor style, giving processor architects the power to live and create tradeoffs in micro architectural selections. for instance, if a benchmark extracts the key algorithms of associate application, it'll contain the performance-sensitive aspects of that application. Running this abundant smaller snipping on a cycle-accurate machine will provide clues on a way to improve performance.

3. verbal description's 1st benchmark suite to live cloud performance SPEC Cloud_IaaS 2016's use is targeted at cloud suppliers, cloud shoppers, hardware vendors, virtualization package vendors, application package vendors, and educational researchers. The benchmark addresses the performance of infrastructure-as-a-service   public or personal cloud platforms. The benchmark is meant to worry provisioning also as runtime aspects of a cloud victimization I/O and central processor intensive cloud computing workloads. verbal description chosen the social media NoSQL info dealings and K-Means agglomeration victimization map/reduce as 2 vital and representative employment sorts inside cloud computing.

The Standard Performance analysis Corporation   may be a non-profit organization that aims to provide honest, impartial and meaning benchmarks for computers. verbal description was supported in 1988 and is supported by its member organizations that embrace all leading pc & package makers. Verbal description benchmarks square measure wide used these days in evaluating the performance of pc systems.

The benchmarks aims to check real-life things , for instance, tests net servers performance by performing arts varied styles of parallel HTTP requests, and SPEC_CPU tests central processor performance by measure the run time of many programs like the compiler gcc and also the chess program foxy. the assorted tasks square measure appointed weights supported their perceived importance; these weights square measure accustomed calculate one benchmark lead to the top.

SPEC benchmarks square measure written in an exceedingly platform neutral artificial language   and also the interested parties might compile the code victimisation no matter compiler they like for his or her platform, however might not modification the code. makers are noted to optimize their compilers to enhance performance of the assorted verbal description benchmarks.